Implementation Of License Plate Recognition Technology Based On Blob Analysis And Joint Feature Extraction

In the rapid development of science and technology and economy today,the increasing number of motor vehicles,increasing the difficulty of management,in order to improve the efficiency of vehicle management,the social from all walks of life in the effective management of urban motor vehicles to make a lot of research.In recent years,the domestic license plate recognition research has also achieved fruitful results,and widely used in toll stations,scenic spots entrance,parking lot,the application technology has become increasingly mature.In the process of intelligent machine recognition,it is difficult to meet the needs of vehicle management in terms of rapid recognition and accurate recognition.Sometimes,human intervention has to be added.Some feature extraction is still dependent on human,which greatly reduces the management efficiency.In some specific complex environments,such as moving vehicles,vehicles under strong light irradiation,vehicles under low light conditions,vehicle license plate defaced skewed vehicles,etc.,the efficiency of intelligent license plate recognition is lower.In recent years,the license plate recognition system based on traditional algorithm is not suitable for high efficiency and accuracy.In this article,the author of more than 2500 vice all kinds of license plates under complicated conditions have done a lot of experiments and research,this paper puts forward a kind of vehicle license plate recognition of image color channel split joint Blob analysis method,to complex background illumination,under the condition of license plate character soiled skew the license for precise positioning and recognition,in this article,the body of the research content mainly includes the following several parts:1.Channel splitting combined with Blob analysis can accurately segment and extract license plate characters.In the process of license plate image preprocessing,the combination of channel resolution and Blob analysis and morphological analysis are used.The experiment shows that the image preprocessing method introduced in this paper can make the location of the license plate more accurate,and the original information will not be lost after the segmentation of the license plate area.At the same time,affine transformation and projection transformation are used to correct the license plate distortion.Through horizontal correction and misalignment correction,it is proved that the corrected license plate character segmentation is easier to extract.2.Use a variety of improved model algorithms to solve the problem of licenseplate character recognition interference.This paper introduces the multi-layer perceptron,convolutional neural network and support vector machine which are widely used in machine learning for image processing direction,and USES four kinds of improved models in convolutional neural network method.In the process of training the model,in order to make the recognition results more reliable,interference information such as distortion and defile was added to the original character area,so as to make the recognition of license plate characters more accurate.3.Simulation experiment under Halcon environment improves the efficiency of license plate character recognition.In the Halcon environment,the license plate recognition was simulated,and the original license plate data were processed and simulated respectively.Through the simulation experiment,it is shown that the license plate recognition has higher recognition rate and better robustness under the background of different complex conditions and the quality of license plate pictures.